Dataset Viewer
The dataset viewer is not available for this subset.
Cannot get the split names for the config 'default' of the dataset.
Exception: SplitsNotFoundError
Message: The split names could not be parsed from the dataset config.
Traceback: Traceback (most recent call last):
File "/usr/local/lib/python3.14/site-packages/datasets/packaged_modules/json/json.py", line 290, in _generate_tables
pa_table = paj.read_json(
io.BytesIO(batch), read_options=paj.ReadOptions(block_size=block_size)
)
File "pyarrow/_json.pyx", line 364, in pyarrow._json.read_json
File "pyarrow/error.pxi", line 155, in pyarrow.lib.pyarrow_internal_check_status
File "pyarrow/error.pxi", line 92, in pyarrow.lib.check_status
raise convert_status(status)
pyarrow.lib.ArrowInvalid: JSON parse error: Column() changed from object to string in row 0
During handling of the above exception, another exception occurred:
Traceback (most recent call last):
File "/usr/local/lib/python3.14/site-packages/datasets/inspect.py", line 286, in get_dataset_config_info
for split_generator in builder._split_generators(
~~~~~~~~~~~~~~~~~~~~~~~~~^
StreamingDownloadManager(base_path=builder.base_path, download_config=download_config)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
)
^
File "/usr/local/lib/python3.14/site-packages/datasets/packaged_modules/json/json.py", line 101, in _split_generators
pa_table = next(iter(self._generate_tables(**splits[0].gen_kwargs, allow_full_read=False)))[1]
~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/usr/local/lib/python3.14/site-packages/datasets/packaged_modules/json/json.py", line 304, in _generate_tables
batch = json_encode_fields_in_json_lines(original_batch, json_field_paths)
File "/usr/local/lib/python3.14/site-packages/datasets/utils/json.py", line 111, in json_encode_fields_in_json_lines
examples = [ujson_loads(line) for line in original_batch.splitlines()]
~~~~~~~~~~~^^^^^^
File "/usr/local/lib/python3.14/site-packages/datasets/utils/json.py", line 20, in ujson_loads
return pd.io.json.ujson_loads(*args, **kwargs)
~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^
ValueError: Expected object or value
The above exception was the direct cause of the following exception:
Traceback (most recent call last):
File "/src/services/worker/src/worker/job_runners/config/split_names.py", line 68, in compute_split_names_from_streaming_response
for split in get_dataset_split_names(
~~~~~~~~~~~~~~~~~~~~~~~^
path=dataset,
^^^^^^^^^^^^^
config_name=config,
^^^^^^^^^^^^^^^^^^^
token=hf_token,
^^^^^^^^^^^^^^^
)
^
File "/usr/local/lib/python3.14/site-packages/datasets/inspect.py", line 340, in get_dataset_split_names
info = get_dataset_config_info(
path,
...<6 lines>...
**config_kwargs,
)
File "/usr/local/lib/python3.14/site-packages/datasets/inspect.py", line 291, in get_dataset_config_info
raise SplitsNotFoundError("The split names could not be parsed from the dataset config.") from err
datasets.inspect.SplitsNotFoundError: The split names could not be parsed from the dataset config.Need help to make the dataset viewer work? Make sure to review how to configure the dataset viewer, and open a discussion for direct support.
YAML Metadata Warning:empty or missing yaml metadata in repo card
Check out the documentation for more information.
CS 546 Homework 2 (2026) submission
Completed RAG notebook and automatically exported results for instructor grading. Use the submitted commit SHA to download the exact submitted version.
What's inside
- Completed notebook with outputs (retrieval, generation, diagnostics, paired challenge)
- Exported function source:
dense_similarity,bm25_similarity,get_topk,hitrate_at_k,prepare_generator_input submission_manifest.json(checksums)
Setup
Edited-fact multiple-choice QA (6 options) over a shared 371-fact corpus; fixed 128-question validation subset (seed 42). Generator: Qwen2.5-3B-Instruct. The retriever is selected automatically by validation Hit@4. Evaluation also uses a 24-question paired challenge on invented two-hop worlds.
Results
| Metric | Value |
|---|---|
| Hit@4 (BM25 / dense) | 0.867 / 0.875 (dense selected) |
| Accuracy at k = 0 / 1 / 4 / 8 | 0.11 / 0.44 / 0.42 / 0.41 |
| Oracle accuracy (32 questions) | 0.44 |
| Paired challenge (mean) | 0.33 |
| Parse errors | 0.0 |
Key findings
- Retrieval depth has diminishing returns. Dense Hit@K rises only from 0.875 (k=4) to 0.883 (k=8), and accuracy falls slightly, since extra facts are mostly distractors.
- Two-hop retrieval fixes the bottleneck (tested separately, not submitted).
A variant that expands the query with the top retrieved document raised
challenge coverage from 9/24 to 22/24 and paired accuracy from 0.33 to 0.83.
The graded
bm25_similaritywas kept to the spec. - Multi-hop questions are the bottleneck. The question names a person but not the country, so single-pass retrieval often misses the second fact. In the challenge, both gold facts were retrieved for only 9 of 24 questions.
- Prompt matters once evidence is present. After prompt revision, the model answered most covered challenge questions correctly; the remaining failures were mainly retrieval misses.
- Downloads last month
- 96